Would it be possible for researchers back-assign a DOI to an existing dataset they have rights to - where there is no permission to share it publically (privacy, organization proprietary, personnel to clean data...etc)?
This would at least allow tracking the same dataset usage across multiple papers (+ - stages of).
To clarify (30 Jan 2019): What I mean is for example there are many papers already published that use the same dataset. In some cases it is not possible to get rights to the dataset. However, at the very least I believe it is useful to back-assign a DOI to these datasets that would allow for people to track datasets across multiple papers, particularly in this age when we may inadvertently commit the sin of deriving effect sizes from the same population if the dataset being the same is not clear, as we sometimes may surmise where people who collected the data get authorship rights directly grandfathered in - but in cases where they are not we may not be able to tell that we are using repeat datasets!
One way to get around the fact that these datasets are not available on open access is to have, under their DOI, descriptions of items that are included, waves of collection, and sample size - these can generally be derived and gradually pieced together from past studies. It then becomes clearer for replication or es. estimation whether the re-analysis actually used the same parts of the dataset. For example using only wave-2 instead of both waves, and using depression from Beck's inventory rather than depression based on interview.